CohereForAI/Global-MMLU-Lite

Global-MMLU-Lite是一个多语言评估数据集，涵盖15种语言，包括英语。它是原始Global-MMLU数据集的“精简”版本，每种语言包含200个文化敏感（CS）和200个文化无关（CA）样本。数据集由专业标注员和Cohere For AI社区的贡献者精心策划，包含多个字段如样本ID、主题、主题类别、问题、选项（a到d）、答案、所需知识、时间敏感性、参考、文化、地区、国家、文化敏感性标签和是否标注的标志。数据集分为测试集和开发集，分别包含6,000和4,275个实例，覆盖15种语言。该数据集遵循Apache 2.0许可证。

Global-MMLU-Lite is a multilingual evaluation dataset spanning 15 languages, including English. It is a lite version of the original Global-MMLU dataset, containing 200 Culturally Sensitive (CS) and 200 Culturally Agnostic (CA) samples per language. The dataset is curated by professional annotators and contributors from the Cohere For AI Community and includes fields such as sample_id, subject, subject_category, question, options (a to d), answer, required_knowledge, time_sensitive, reference, culture, region, country, cultural_sensitivity_label, and is_annotated. The dataset is split into test and dev sets, with 6,000 and 4,275 instances respectively, covering 15 languages. The dataset is licensed under Apache 2.0.